Theory and Engineering of Scheduling Parallel Jobs
Abstract (englisch):
Scheduling is very important for an efficient utilization of modern parallel computing systems. In this thesis, four main research areas for scheduling are investigated: the interplay and distribution of decision makers, the efficient schedule computation, efficient scheduling for the memory hierarchy and energy-efficiency. The main result is a provably fast and efficient scheduling algorithm for malleable jobs. Experiments show the importance and possibilities of scheduling considering the memory hierarchy.
| Zugehörige Institution(en) am KIT | Institut für Theoretische Informatik (ITI) |
| Publikationstyp | Hochschulschrift |
| Publikationsjahr | 2018 |
| Sprache | Englisch |
| Identifikator | urn:nbn:de:swb:90-841531 KITopen-ID: 1000084153 |
| Verlag | Karlsruher Institut für Technologie (KIT) |
| Umfang | 271 S. |
| Art der Arbeit | Dissertation |
| Fakultät | Fakultät für Informatik (INFORMATIK) |
| Institut | Institut für Theoretische Informatik (ITI) |
| Prüfungsdatum | 14.12.2017 |
| Schlagwörter | scheduling, malleable jobs, parallel computation, scheduling for the memory hierarchy, scheduling for energy efficiency |
| Referent/Betreuer | Sanders, P. |